Opinion Mining Using Supervised Machine Learning Technique to Monitor User Reviews Over Social Networking Applications

Abstract

In recent trends, Opinion Mining (OM) and Sentimental Analysis (SA) are becoming popular on newlinethe internet and are utilized in many applications from small to large commercial environments. newlineBefore making a purchase, individuals used to solicit input from friends and family members. newlineWith the advent of social media and microblogging, peoples come to know what others are newlinethinking about a service or about a product before they making a purchase. Pre-processing and newlinecategorizing tweets from a Twitter location into advantageous, irrelevant, and unfavourable newlinecategories are some of the methodologies that are examined in this thesis. A number of newlinecategorization techniques based on recall and accuracy are also being investigated. OM is one of newlinethe methods for analysing, classifying, extracting and interpreting the opinions expressed by newlinevarious persons. Recently, sentiment classification is analysed with the majority of the problem newlinefocusing on classifying blog posts and movie critique, service feedbacks etc., newlineMethods for classifying and relating current techniques are presented in this thesis. Two newlinealternative clustering and classification algorithms were used in the first study. A robust newlinehierarchical clustering method (ROCK) is used to build the clusters, and the CART algorithm newlineclassifies the words as positive or negative. After everything is said and done, use gets a newlinerecommendation on the movie which has the greatest percentage of positive viewers feedback, newlinewhich is then classified and overall accuracy of user s remarks are evaluated. In order to complete newlinecustomer evaluations for different movie sectors, this analysis helps. Accordingly, the suggested newlineapproach has an accuracy rate of 89.76%. Aspect-based sentiment classification was tested in the newlinesecond experiment as a method for evaluating user sentiments. Pre-processing truncates and newlineremoves everything except the URL, stop words, and emoticons and symbols from tweets. To newlineidentify and extract important data from cleaned and processed tweets, two major feature newlineextraction algori